Author: Bob Durrett
Date: 15:00:15 01/15/04
Given that a game has already taken t seconds and n half-moves, it seems one could estimate the time or number of remaining moves [or total time or number of half-moves] which are expected. One might be more specific by asking "what is the amount of ramaining time which has 50% probability of occurring" or "what is the number of remaining moves which has a 50% probability of occurring. This is similar to questions about a human's life expectancy. In certain countries, life expectancy at birth is about 80 years for a female. However, if it is already know that a specific female in that country has reached the age of 46 years, then the apriori life expectancy for that individual [and for all other 46 year old females in that country] would be somewhere between 80 and 100. The relevance of this to chess engines is in the area of time management. It has been said that a properly designed chess-playing program should never lose on the clock. I do not know [and cannot know] how commercial chess engines do their time management. [In view of Dann Corbett's list, there are also too many amateur engines to know about too.] Intuitively, it would seem that a chess-playing program would do such calculations to determine the optimum amount of time to take in calculation of the next move. [This time must take into account time used by anything external to the engine.] My question is: Do most chess engines do their time management this way? If not, how is what they do different from the above? Bob D.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.